What Moonlux observed today
Rates and requirements come from the publishers. Cards are condensed and may omit constraints such as work authorization; confirm every requirement in the original listing. A listing is an opportunity to apply, not a promise of work. See the full board and methodology.
Where are you right now?
Choose the closest answer. The result stays in this tab; Moonlux records only the route selected, a temporary page-session key, and campaign source so we can learn which part of the hub lawyers need. We do not receive answer text, identity, or work content.
Four doors into the work
1. Find a fitting opening
See current roles Moonlux has observed, published pay, stated requirements, and what a talent-network listing actually means.
Current legal AI jobs →2. Understand the task
Learn the Build, Break, Answer, Grade loop—and why project instructions and rubrics matter more than generic prompting tricks.
How the work works →3. Build a coherent world
Map the six-stage workflow, protect frozen facts, and prevent one change from breaking the rest of the corpus.
Open the phase planner →4. Reduce preventable rework
Use a practical pre-submission checklist for scope, files, internal consistency, metadata and checks that did not actually run.
Use the checklist →The opportunity and the friction are both real
Expert judgment is the product
Current public listings ask lawyers to evaluate reasoning, verify authority, create realistic scenarios, draft reference work and write rubrics—not merely label text.
Matching comes before mastery
Our earliest practitioner research says a strong lawyer may still wait for a project match. Profiles, assessments, jurisdiction, specialty and current demand all shape access.
Instructions change by project
Tool permissions, required artifacts, evaluation standards and submission mechanics are project-specific. A practice that is allowed in one phase may be barred in another.
Economics depend on the project
On hourly work, a QC cycle may still be paid. With task bonuses or fixed rewards, cycle time affects earnings directly. Either way, preventable rework consumes capacity and attention.
Moonlux’s early field research includes three recorded conversations with one practicing attorney doing legal AI work and a separate walkthrough of a world-building project. Those interviews give us unusually concrete workflow detail, but they are still one contributor’s experience. We use them to form and test hypotheses—not to claim every platform or project works the same way.
Build. Break. Answer. Grade.
| Move | What the lawyer contributes | Typical failure |
|---|---|---|
| Build | Create a realistic legal problem and the documents needed to solve it. | The scenario contradicts itself, leaks the answer, or does not match the deliverable. |
| Break | Design ambiguity, competing authority or factual tension that reveals weak reasoning. | The task tests obscure retrieval rather than judgment—or accidentally becomes easy. |
| Answer | Produce the reference analysis or work product at professional standard. | The conclusion is right but the reasoning, support, scope or format is not. |
| Grade | Write criteria that separate legally meaningful errors from style preferences. | The rubric double-counts one issue, rewards observation instead of reasoning, or cannot be applied consistently. |
What Moonlux does—and does not do
Moonlux tracks public legal AI openings, helps attorneys understand which route may fit, and is building practical support for the work itself. Merit by Moonlux checks selected files privately in a modern desktop browser for defined mechanical and workflow problems before submission.
Moonlux is independent. We are not Mercor, Handshake, DataAnnotation, Outlier, an AI lab, or your employer. The platforms contract with and pay contributors directly. Merit does not replace the lawyer’s substantive review and does not promise platform acceptance or steady work.
Short answers
Do I need an AI background?
Most current legal-expert listings emphasize legal experience, research, judgment and clear writing. Some explicitly say prior AI experience is not required. You still need to learn each project’s tools and rules.
Is this prompt engineering?
Usually not in the popular sense. The harder work is constructing and evaluating legal reasoning: choosing the right facts and authorities, spotting unsupported conclusions, creating reference work, and writing a usable rubric.
Is the work available outside the United States?
Some roles accept lawyers in several countries and some are jurisdiction-specific. Read the listing itself; location, work authorization and licensing requirements change by role.
Can Moonlux guarantee a match?
No. Moonlux can make opportunities easier to find and a candidate easier to understand, but the platform and project determine matching and selection.
The Legal AI Work Starter Kit
A one-page field guide for finding a credible opening, preparing for an assessment and getting a first task through a clean final pass. It is free, public and designed to print or share.
Open the starter kitWant Moonlux in your corner?
Join the founding cohort. Tell us what you practice, what platforms you already use and where you are licensed. We will use that information to surface relevant work and improve Merit around the tasks lawyers actually receive.
Join MoonluxTry Merit first